Multicentre Evaluation of the EUCAST Rapid Antimicrobial Susceptibility Testing (RAST) Extending Analysis to 16–20 Hours Reading Time

The aim of the study was to evaluate the EUCAST RAST method by extending analysis to 16–20 h reading time and performance with new β-lactam/β-lactamase inhibitor combinations. A total of 676 positive blood cultures (BCs) were enrolled. Results at 4 h, 6 h, 8 h and 16–20 h were interpreted according to bacterial species using EUCAST RAST breakpoints (version 5.1). For species for which no breakpoints were available, tentative breakpoints were used. Categorical agreement with the Microscan microdilution system was analysed. Among the 676 BCs enrolled, 641 were monomicrobial and were included in the analysis. Categorical agreement ranged from 98.9% at 4 h to 99.4% at 16–20 h. The rates of very major errors were 3.3%, 3.7% and 3.4% at 4 h, 6 h and 8 h, respectively, and decreased to 1% at 16–20 h (p < 0.001). The number of major errors was low for each reading time (0.2% and 0.4% at 4 h and 6 h, respectively, and 0.3% at both 8 h and 16–20 h). The proportions of results in the area of technical uncertainty were 9.9%, 5.9%, 5% and 5.2% for readings at 4 h, 6 h, 8 h and 16–20 h, respectively. Tentative breakpoints proposed for Enterobacterales other than E.coli/K.pneumoniae and coagulase-negative staphylococci showed overall performances comparable to those observed for E. coli/K. pneumoniae and S. aureus. In conclusion, EUCAST RAST has been shown to be reliable to determine microbial susceptibility to main antimicrobials, including ceftazidime/avibactam and ceftolozane/tazobactam. A poorer performance was observed for certain species/antimicrobial agent combinations. The better performance observed at 16–20 h compared to the early readings may confer to the method greater potential for antimicrobial de-escalation interventions.


Introduction
Bloodstream infections (BSIs) are a major cause of morbidity and mortality with increasing incidence worldwide, often due to delayed or inappropriate antimicrobial treatment [1][2][3]. Hence, blood cultures (BCs) remain essential to identifying causative pathogens and their antimicrobial susceptibility profiles. Several molecular-based systems and MALDI-TOF MS approaches have been implemented directly from positive BCs [4,5]. However, given the wide range of antibiotic-resistance mechanisms and their increasing spread worldwide, rapid antimicrobial susceptibility tests (RASTs) have shown potential for shortening time to effective treatment and consequently improving clinical outcomes in BSIs [6][7][8]. Recently, EUCAST developed a phenotypic RAST method based on disc diffusion and performed directly from positive BC bottles to provide results after 4 h, 6 h and 8 h of incubation [9][10][11]. The method has currently been validated for four
MEs were found only for piperacillin/tazobactam in E. coli (2.4% at 4 h and 6 h) and K. pneumoniae (1.6% at 6 h). Of note, no VMEs or MEs were observed with readings at 16-20 h.
CA was 98.1% and 95.9% in S. aureus and CoNS, respectively (Table 5). Overall ATU results were significantly higher in CoNS than S. aureus isolates (10% vs. 1.7%, respectively) (p < 0.001). The RAST test for screening of clindamycin-inducible resistance was essential to detect clindamycin-inducible resistance, allowing the identification of 42.4% (n = 14) and 52.3% (n = 23) of the total clindamycin-resistant S. aureus (n = 33) and CoNS isolates (n = 44), respectively. However, VMEs for clindamycin were found at 4 h, 6 h and 8 h in S. aureus (ranging from 6.1% to 8.3%) and CoNS isolates (ranging from 2.3% to 4.5%), whereas MEs were only observed in CoNS (range: 0-3.6%). Similarly, VMEs were observed for gentamycin at 4 h, 6 h and 8 h but not at 16-20 h. Excluding ATU results and nonreadable zones, cefoxitin RAST screening identified all methicillin-resistant staphylococci at each reading time. Abbreviations: S, susceptible; I, susceptible at increased exposure; R, resistant; ATU, area of technical uncertainty; VME, very major error; ME, major error; mE, minor error.

E. faecalis and E. faecium
RAST overall categorical agreement was 100% and 97.9% in E. faecalis and E. faecium, respectively (Table 6). Excluding vancomycin, for which the breakpoint of susceptibility was not available, ATU results were mainly found for linezolid in both species (3.2-11.1% in E. faecalis, 18.7-100% in E. faecium) and for gentamicin in E. faecalis (12.5-50%). One VME and one ME were observed in E. faecium for ampicillin at 4 h, and for linezolid at 6 h, respectively. A lower number of ATU results and no categorical errors were found with readings at 16-20 h. Abbreviations: S, susceptible; I, susceptible at increased exposure; R, resistant; ATU, area of technical uncertainty; VME, very major error; ME, major error; mE, minor error.

Discussion
Although several RAST methods for BCs have been developed over time, their use has been confined to experimental studies or limited hospital settings [17][18][19][20]. Since the end of 2018, the availability of species-specific EUCAST-validated RAST breakpoints for the interpretation of inhibition zones at 4 h, 6 h and 8 h has marked a turning point in the implementation of EUCAST RAST in the BC routine of most microbiology laboratories.
In this study, EUCAST RAST showed excellent CA with automated microdilution Microscan system. However, an overall rate > 3% of false susceptibility results (VMEs) was observed, mainly due to a poor performance of the RAST for certain species/antimicrobial agent combinations. Furthermore, the limited number of isolates resistant to some of the antibiotics tested could represent an additional bias. As also shown in previous published data [13,[21][22][23], VMEs were mainly observed: in aminoglycosides for E. coli, K. pneumoniae and other-EB; in cephalosporins for E. coli and other-EB; in gentamicin and clindamycin for S. aureus and CoNS. EUCAST RAST was shown to perform better for P. aeruginosa and A. baumannii (no VMEs) and for enterococci (only one false susceptible result).
Tentative breakpoints proposed for other-EB and CoNS showed satisfactory performances, with overall CA and error rates almost comparable to those obtained for E. coli/K. pneumoniae and S. aureus, respectively. However, high error rates were obtained for Proteus mirabilis isolates, suggesting that further species-specific EUCAST RAST breakpoints are desirable. In addition, a higher number of results falling in the ATU was obtained for both CoNS and other-EB. In the latter case, the explanation could be the setting of tentative breakpoints, characterized by a wider ATU aimed at minimising categorisation discrepancies.
Implementation of the RAST method for new β-lactams/β-lactamase inhibitors could be essential in geographical areas with endemic diffusion of MDR Gram-negative organisms, such as carbapenemase-resistant Enterobacterales and P. aeruginosa [24,25]. However, data on RAST accuracy for these antimicrobial combinations are lacking. We recently showed a good performance of EUCAST RAST in the determination of ceftazidime/ avibactam susceptibility for carbapenemase-producing Enterobacterales, including both metallo-β-lactamase and serine-carbapenemase producers [16]. The present study confirmed these findings, since EUCAST RAST was demonstrated to be reliable, identifying all four ceftazidime/avibactam-resistant isolates and giving only a false resistant result for a P. aeruginosa isolate. However, further studies involving populations with higher rates of resistant isolates are warranted to extend these findings. To date, no study has investigated the performance of RAST in ceftolozane/tazobactam, a combination mainly used for the treatment of MDR P. aeruginosa and ESBL-producing Enterobacterales. In this study, 97.2-100% CA was shown for K. pneumoniae, E. coli, other-EB and P. aeruginosa, with no VMEs and only one ME at 6 h for a K. pneumoniae isolate.
Concerning the evaluation of the EUCAST RAST breakpoints for incubation at 16-20 h, we observed an improvement in performance, with higher rates of CA, reduction in the frequency of VMEs and a number of MEs almost similar to that observed at 8 h. Another advantage found with readings at 16-20 h was a strong reduction in the percentage of results falling within the ATU for piperacillin/tazobactam in E. coli, K. pneumoniae and other-EB.
De-escalation and escalation of empirical antibiotic therapy guided by RAST results is the goal for which this rapid method was developed [13,[21][22][23]. However, the limited data available on this topic may, at least at present, limit the use of RAST susceptibility results to perform de-escalation interventions [13]. The delayed reading of the inhibition zones at 16-20 h, which is a very close timing to that of the standard disc-diffusion method, together with the lower rate of false susceptible results at 16-20 h compared to early readings, could strengthen the relevance of the RAST results at 16-20 h for de-escalation interventions.

Materials and Methods
The study was carried out in two tertiary-care Italian teaching hospitals ("Città della Salute e della Scienza di Torino", Turin, and "Sant' Orsola-Malpighi", Bologna). During a six-month period, positive BCs were processed according to the diagnostic protocols of each laboratory (Table 7). A total of 676 positive BCs, which showed Gram-negative rods or Gram-positive cocci at Gram-staining examination, were enrolled to evaluate EUCAST RAST. EUCAST RAST was performed according to EUCAST guidelines within 2 h after the removal of a positive bottle from a BC incubator [9]. According to Gram staining and morphological classification, the predetermined antibiotic discs were placed on Mueller-Hinton plates immediately after inoculation and spreading of 150 µL of the BC fluid (Table  S1). All plates were incubated at 35 ± 1 • C in ambient air. Inhibition zones were read at 4 h, 6 h and, when possible, at 8 h and 16-20 h. The antibiotic-disc manufacturer, brand of Mueller-Hinton medium and mode of measurement of inhibition zones are reported in Table 7. According to EUCAST recommendations, the quality-control procedure was performed by both laboratories to validate the performance of antibiotic-discs, the agar used and the methods of reading inhibition zones using E. coli ATCC 25922, P. aeruginosa ATCC 27853, S. aureus ATCC 29213 and E. faecalis ATCC 29212 [26]. Interpretations of inhibition zone diameters were carried out in accordance with EU-CAST RAST breakpoints (version 5.1) [12], and the results were compared to those obtained by reference susceptibility testing to delineate CA [27]. Polymicrobial BCs were excluded from the analysis. For other-EB, a breakpoint table was obtained by combining the RAST breakpoints for E. coli and K. pneumoniae, considering the largest values of 'susceptible' and 'resistant' (Table S2). Since EUCAST RAST guidelines do not include breakpoints for CoNS, the RAST results were determined according to S. aureus RAST breakpoints. Categorical errors were classified as VME (susceptible on RAST, but resistant on reference AST), ME (resistant on RAST, but susceptible on reference AST), and minor error (mE) (susceptible or resistant on RAST, but susceptible requiring increased antibiotic exposure on reference AST). To this end, isolate/drug combinations falling within ATU or considered uninterpretable at the time of reading (i.e., insufficient growth) were excluded. Since EUCAST classifies piperacillin/tazobactam, ceftazidime, cefepime, imipenem and levofloxacin as "susceptible, but requiring increased antibiotic exposure" and not as "susceptible" for P. aeruginosa, discrepancies herein were considered major rather than minor errors (i.e., if resistant according to RAST but susceptible requiring increased antibiotic exposure by reference AST). The same rationale was followed for imipenem in Morganellaceae (Morganella morganii, Proteus spp., Providencia spp.) and in Enterococcus spp.
Comparisons involving dichotomous variables were tested using the X 2 test. Statistical significance was set at a p-value < 0.05.

Conclusions
Our results showed that the RAST method is reliable for determining microbial susceptibility to main antimicrobials, including ceftazidime/avibactam and ceftolozane/ tazobactam. A poorer performance was observed for certain species/antimicrobial agent combinations. The use of tentative RAST breakpoints for other-EB and CoNS has been shown to be feasible while awaiting further validated species-specific breakpoints. Readings at 16-20 h make the EUCAST RAST method applicable regardless of laboratory opening hours. Additionally, EUCAST RAST showed better performance at 16-20 h compared to early readings, giving it greater potential for antimicrobial de-escalation interventions.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics11101404/s1. Table S1: Antibiotic discs used for EU-CAST RAST according to microscopic examination results; Table S2: Tentative zone diameter RAST breakpoints for Enterobacterales other than E. coli/K. pneumoniae.

Data Availability Statement:
The authors confirm that the data supporting the findings of this study are available from the corresponding author on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.